Research map / Label-efficient and robust learning / Recognition and long-tail
Tokenize Once, Recommend Anywhere: Unified Item Tokenization for Multi-domain LLM-based Recommendation
Yu Hou, Won-Yong Shin
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add TokenMoE domain-specific experts (ablation) | UniTok tokenizer · Beauty (ablation) | NDCG@10: 0.0304 -> 0.0436 | Improved |
| add the always-active shared expert (ablation) | UniTok tokenizer · Beauty (ablation) | NDCG@10: 0.0436 -> 0.0457 | Improved |
| add MI calibration loss (ablation) | UniTok tokenizer · Beauty (ablation) | NDCG@10: 0.0457 -> 0.0478 | Improved |
| add TokenMoE domain-specific experts (ablation) | UniTok tokenizer · Cellphones (ablation) | NDCG@10: 0.0371 -> 0.0606 | Improved |
| add the always-active shared expert (ablation) | UniTok tokenizer · Cellphones (ablation) | NDCG@10: 0.0606 -> 0.0622 | Improved |
| add MI calibration loss (ablation) | UniTok tokenizer · Cellphones (ablation) | NDCG@10: 0.0622 -> 0.0647 | Improved |
| add TokenMoE domain-specific experts (ablation) | UniTok tokenizer · Grocery (ablation) | NDCG@10: 0.0342 -> 0.0497 | Improved |
| add the always-active shared expert (ablation) | UniTok tokenizer · Grocery (ablation) | NDCG@10: 0.0497 -> 0.0515 | Improved |
| add MI calibration loss (ablation) | UniTok tokenizer · Grocery (ablation) | NDCG@10: 0.0515 -> 0.0533 | Improved |
| replace separate per-domain tokenizers with one unified UniTok model | codebook tokenizers · 10 datasets | trainable parameters (M): 87.78M -> 9.11M | Improved |
| use UniTok instead of LETTER when jointly trained on ten datasets | LLM recommender tokenizer · Beauty (single unified training setup) | NDCG@10: 0.0288 -> 0.0478 | Improved |
| use UniTok instead of LETTER zero-shot | LLM recommender tokenizer · Health (zero-shot) | NDCG@10: 0.0375 -> 0.0442 | Improved |
Closest papers by meaning
- Semantic Convergence: Harmonizing Recommender Systems via Two-Stage Alignment and Behavioral Semantic Tokenization (AAAI 2025 · similarity 0.90)
- Improving LLMs for Recommendation with Out-Of-Vocabulary Tokens (ICML 2025 · similarity 0.90)
- Can LLMs Outshine Conventional Recommenders? A Comparative Evaluation (NeurIPS 2025 · similarity 0.88)
- LLM-ESR: Large Language Models Enhancement for Long-tailed Sequential Recommendation (NeurIPS 2024 · similarity 0.87)
- LLMEmb: Large Language Model Can Be a Good Embedding Generator for Sequential Recommendation (AAAI 2025 · similarity 0.87)
- LLM4RSR: Large Language Models as Data Correctors for Robust Sequential Recommendation (AAAI 2025 · similarity 0.86)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
